Paper Title
Improving Pneumonia Classification Using KNN and Random Forest: A Machine Learning Perspective

Abstract
Evaluating the efficacy and efficiency of neural network training for pneumonia detection by comparing K-Nearest Neighbors (KNN) and Random Forest. This study examines their impact on robustness and classification accuracy, providing insights into optimal method selection for medical image analysis. A diverse set of categorized chest X-ray images from Kaggle is used for model training and evaluation. Both KNN and Random Forest are implemented, with performance metrics optimized to maximize classification accuracy. Data augmentation techniques and a threefold cross-validation approach are applied to enhance robustness. Experimental results indicate that KNN outperforms Random Forest, achieving an accuracy of 98.13%, while Random Forest attains 97.28%. These findings suggest that KNN is particularly well-suited for pneumonia detection in medical imaging, though both models exhibit strong potential in disease classification. Further investigation is required to understand the factors influencing performance differences and to explore additional refinements for improving diagnostic accuracy. This study highlights the importance of algorithm selection in neural network-based medical image processing and the need for continued advancements in deep learning techniques for pneumonia detection. Keywords - Pneumonia detection, KNN, Random Forest, Chest X-ray images, Data augmentation, Classification accuracy.